Loading npm security reports…

audit-tools@0.50.3: AI-agent control-surface risk | LPM Firewall

registry / audit-tools / 0.50.3

audit-tools@0.50.3

Provider-neutral code auditing and remediation workflows for arbitrary repositories.

AI Security Review

scanned 3h ago · by lpm-firewall-ai

LPM flags this version as an AI-agent control-surface risk. Installing the package automatically deploys global AI-agent commands, skills, plugin files, and OpenCode configuration. These writes affect several host applications outside the installed project.

Static reason

High-risk behavior combination matched malicious policy.; source matched previously finalized malicious package; routed for review

Trigger

npm postinstall during package installation

Impact

A package install changes Claude, Codex, OpenCode, and Antigravity behavior without a separate user setup command.

Mechanism

Automatic global AI-agent control-surface mutation

Policy narrative

On installation, the postinstall entrypoint launches both deployers. They write package-provided prompts and skills into global Claude and Codex directories, merge global OpenCode commands and agent permissions, and add Antigravity plugins. The write helper creates directories and replaces prior generated files. This is unconsented install-time mutation of broad, foreign AI-agent control surfaces.

Rationale

The package performs confirmed postinstall writes to multiple global AI-agent configuration locations. This meets the blocking policy regardless of the absence of confirmed credential exfiltration.

Evidence

package.jsonscripts/postinstall.mjsscripts/audit/postinstall.mjsscripts/remediate/postinstall.mjsscripts/shared/install-host-assets.mjs~/.claude/commands/audit-code.md~/.claude/commands/remediate-code.md~/.claude/plugins/marketplaces/claude-plugins-official/external_plugins/audit-code

Source & flagged code

7 flagged · loading source

package.jsonView file

•scripts.postinstall = node scripts/postinstall.mjs

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg

•scripts.postinstall = node scripts/postinstall.mjs

Medium

Ambiguous Install Lifecycle Script

Install-time lifecycle script is not statically allowlisted and needs review.

package.jsonView on unpkg

wrapper/audit-code-wrapper-lib.mjsView file

340const distUrl = new URL(`file:///${distCliEntry.replace(/\\/g, '/')}`); L341: const cli = await import(distUrl.href); L342: await cli.runCli([process.execPath, distCliEntry, commandName, ...commandArgs]);

Medium

Dynamic Require

Package source references dynamic require/import behavior.

wrapper/audit-code-wrapper-lib.mjsView on unpkg · L340

dist/audit/orchestrator/reviewPackets.jsView file

1import { createHash } from "node:crypto"; L2: import { collectGraphEdges, compareCodeUnits } from "audit-tools/shared";

Low

Weak Crypto

Package source references weak cryptographic algorithms.

dist/audit/orchestrator/reviewPackets.jsView on unpkg · L1

scripts/audit/postinstall.mjsView file

1Install-time code directly mutates a foreign AI-agent control surface: L159: grep: 'allow', L160: edit: { ...OPENCODE_AUDIT_EDIT_PERMISSION }, L161: bash: { ...OPENCODE_AUDIT_BASH_PERMISSION }, L162: }; ... L204: function claudePluginExternalDir() { L205: return join(homedir(), '.claude', 'plugins', 'marketplaces', 'claude-plugins-official', 'external_plugins', 'audit-code'); L206: } ... L222: label: 'Claude command', L223: path: join(homedir(), '.claude', 'commands', 'audit-code.md'), L224: sourcePath: promptSourceFile, ... L228: label: 'Codex skill', L229: path: join(homedir(), '.codex', 'skills', 'audit-code', 'SKILL.md'), Write operation from scripts/shared/install-host-assets.mjs: L12: import { dirname, join } from "node:path"; L13: import { mkdirSync, existsSync, readFileSync, writeF

Critical

Ai Agent Control Hijack

Source creates an unconsented AI-agent control surface through install-time mutation or a default unauthenticated remote skill channel.

scripts/audit/postinstall.mjsView on unpkg · L1

dist/audit/cli/args.jsView file

•stage = ast_semantic_analysis; reason = ast_alias_growth_limit_exceeded; limitedFiles = 1

High

Semantic Analysis Limited

A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.

dist/audit/cli/args.jsView on unpkg

wrapper/audit-code-wrapper-build.mjsView file

•matchType = normalized_sha256 matchedPackage = audit-tools@0.45.0 matchedPath = wrapper/audit-code-wrapper-build.mjs matchedIdentity = npm:YXVkaXQtdG9vbHM:0.45.0 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

wrapper/audit-code-wrapper-build.mjsView on unpkg

Source & flagged code

7 flagged

package.jsonView file

Lines 22-62json

22 },

23 "files": [

24 "dist/**",

25 "audit-code.mjs",

26 "wrapper/**",

27 "remediate-code.mjs",

28 "dispatch/**",

29 "schemas/**",

30 "skills/**",

31 "scripts/postinstall.mjs",

32 "scripts/audit/postinstall.mjs",

33 "scripts/remediate/postinstall.mjs",

34 "scripts/shared/install-host-assets.mjs",

35 "opencode.json",

36 "README.md",

37 "docs/audit-pkg/product.md",

38 "docs/audit-pkg/operator-guide.md",

39 "docs/audit-pkg/contracts.md"

40 ],

41 "scripts": {

42 "postinstall": "node scripts/postinstall.mjs",

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg · L42

Medium

Ambiguous Install Lifecycle Script

Install-time lifecycle script is not statically allowlisted and needs review.

package.jsonView on unpkg · L42

43 "prebuild": "node scripts/shared/guard-no-suite-running.mjs",

44 "build": "node scripts/shared/clean-dist.mjs && tsc -p tsconfig.json",

45 "check": "tsc -p tsconfig.json --noEmit",

46 "check:tests": "tsc -p tsconfig.test.json",

47 "check:scripts": "tsc -p tsconfig.scripts.json",

48 "nightly:inbox": "node scripts/nightly/render-inbox.mjs",

49 "nightly:ingest": "node scripts/nightly/ingest-answers.mjs",

50 "test:doc-contract": "node scripts/shared/run-vitest-gate.mjs tests/audit/release-contract.test.ts tests/audit/file-lock-doc-sync.test.ts tests/audit/host-asset-renderer-drift.test.ts",

51 "test": "npm run build && node scripts/shared/run-vitest-gate.mjs",

52 "test:single": "npm run build && node scripts/shared/run-vitest-gate.mjs",

53 "verify:guards": "node scripts/shared/run-vitest-gate.mjs --retry=2 --exclude \"**/audit-code-completion-*.test.ts\" --exclude \"**/audit-code-wrapper-*.test.ts\" --exclude \"**/audit/next-step-core-*.test.ts\" --exclude \"**/seam-host-only-next-step.test.ts\" --exclude \"**/*-e2e.test.*\" --exclude \"**/next-step-implement-dispatch.test.ts\" --exclude \"**/next-step-pipeline-dispatch.test.ts\""

54 "verify:checks": "node scripts/shared/profile-run.mjs verify-checks check:control-bytes check:shared-primitives check:version-gates check:guard-reach check:ci-trigger-paths check:offload-lanes check:loop-core-patterns check:constitutional-doc-paths check:runtime-artifact-names check:executor-producers check:spec-mirrors check:cli-surface check:deadcode check:lint check:dup check:depgraph check:doc-manifest check:doc-links check:doc-code-citations check:gate-enumeration check:philosophy-brief check:readme-sample-report check:proposal-red-at check:nightly-routine-prompt check:handoff-roadmap ...

55 "verify:release": "npm run verify:checks && node scripts/shared/run-vitest-gate.mjs && npm run smoke:linked-audit-code && npm run smoke:linked-remediate-code",

56 "check:control-bytes": "node scripts/check-control-bytes.mjs",

57 "check:shared-primitives": "node scripts/check-shared-primitives.mjs",

58 "check:guard-reach": "node scripts/check-guard-reach.mjs",

59 "closeout": "node scripts/render-closeout.mjs",

60 "check:ci-trigger-paths": "node scripts/shared/generate-ci-trigger-paths.mjs --check",

61 "check:deadcode": "knip --no-config-hints",

62 "check:lint": "eslint .",

Long lines were clipped for display.

wrapper/audit-code-wrapper-lib.mjsView file

Lines 320-360javascript

320 // Gate the mkdir behind the same ensureArtifactsDir flag as runDistCommand so

321 // "the artifacts directory is created only for designated stateful commands"

322 // holds on this path too (CE-001); mcp is a designated stateful command and

323 // opts in explicitly at its call site.

324 if (ensureArtifactsDir) {

325 await mkdir(artifactsDir, { recursive: true

dist/audit/orchestrator/reviewPackets.jsView file

Lines 1-21javascript

1import { createHash } from "node:crypto";

2import { collectGraphEdges, compareCodeUnits } from "audit-tools/shared";

Low

Weak Crypto

Package source references weak cryptographic algorithms.

dist/audit/orchestrator/reviewPackets.jsView on unpkg · L1

3import { priorityRank, sortLenses } from "./auditTaskUtils.js";

scripts/audit/postinstall.mjsView file

Lines 1-21javascript

1#!/usr/bin/env node

2import { homedir } from 'os';

3import { join, dirname } from 'path';

4import { readFileSync } from 'fs';

5import { fileURLToPath } from

dist/audit/cli/args.jsView file

•stage = ast_semantic_analysis; reason = ast_alias_growth_limit_exceeded; limitedFiles = 1

High

Semantic Analysis Limited

A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.

dist/audit/cli/args.jsView on unpkg

wrapper/audit-code-wrapper-build.mjsView file

•matchType = normalized_sha256 matchedPackage = audit-tools@0.45.0 matchedPath = wrapper/audit-code-wrapper-build.mjs matchedIdentity = npm:YXVkaXQtdG9vbHM:0.45.0 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

wrapper/audit-code-wrapper-build.mjsView on unpkg